How can I count CSV records while ignoring blank rows?

Open the file with newline='', pass it to csv.reader, and count each nonempty parsed row. For example:

import csv

count = 0
with open("data.csv", newline="") as f:
    for row in csv.reader(f):
        if row:
            count += 1

print(count)

Here, if row skips a row only when the reader returns an empty list. The standard reader skips a blank physical line; a line containing a delimiter can instead parse as fields containing empty strings. CSV fields can also contain newlines, so counting physical lines does not necessarily count records. [S1]

This counts every nonempty parsed record, including a header if the file has one. If you want data records only, decide separately whether to exclude the header.

A blank physical line is skipped by the standard csv.reader. By contrast, a line containing a comma has a delimiter, so it can produce a parsed row with two empty-string fields. Because that row is still a list with entries, if row keeps it. [S1]

For a simplified illustration, imagine these lines in a CSV file:

name,city

,
Mina,

The blank line contributes no parsed row under the standard reader. The line containing just , represents two empty fields and is retained by the count. Mina, also represents a row and is retained, even though its second field is empty. This approach does not mean “skip any record with missing values”; it means “skip only an empty parsed row.”

That expression counts nonblank physical lines, not parsed CSV records. A quoted field can contain a newline, so a single CSV record may span multiple physical lines. Counting lines could therefore give a different result from iterating over parsed rows. [S1]

It can also blur distinctions you may care about: a line containing commas is not whitespace-only, even if all its fields are empty. To count according to CSV structure, let the reader parse the file first, then apply the empty-row condition to each parsed result.

The example assumes the file is intended to use the dialect selected by csv.reader’s defaults. If the file uses a different delimiter or quoting convention, configure the reader for that dialect; parsing choices affect what counts as a row. [S1]

Suppose the file contains a header, a blank line, a two-empty-field row, and a record whose quoted field spans two physical lines:

name,note

,
Mina,"first line
second line"

Using the standard reader and counting nonempty rows gives three parsed records: the header, the row represented by ,, and Mina’s record. The blank physical line is skipped. Mina’s entry is one parsed record even though its quoted field occupies two physical lines. [S1]

To count data records but not the header, skip the first nonempty parsed row:

import csv

count = 0
header_seen = False
with open("data.csv", newline="") as f:
    for row in csv.reader(f):
        if not row:
            continue
        if not header_seen:
            header_seen = True
            continue
        count += 1

print(count)

This treats the first nonempty parsed row as the header; it does not verify that the row contains particular column names. The example explains standard-reader behavior, not every possible CSV dialect or malformed-file case. [S1]